![[2510.06940] Revisiting Node Affinity Prediction in Temporal Graphs](https://arxiv.org/static/browse/0.3.4/images/arxiv-logo-fb.png){kind=logo}
[2510.06940] Revisiting Node Affinity Prediction in Temporal Graphs
Summary
The paper presents NAViS, a novel model for node affinity prediction in temporal graphs, addressing challenges in current methods and outperforming existing heuristics.
Why It Matters
Node affinity prediction is crucial in various applications, including social networks and recommender systems. This research enhances predictive accuracy, potentially leading to better decision-making in these fields. By introducing NAViS and a new loss function, the authors contribute to advancing machine learning techniques in temporal graph analysis.
Key Takeaways
- NAViS model improves node affinity prediction in temporal graphs.
- Current state-of-the-art models are outperformed by simple heuristics.
- A novel loss function is introduced for better training of NAViS.
- The research addresses significant challenges in training temporal graph neural networks.
- Source code for NAViS is publicly available for further research.
Computer Science > Machine Learning arXiv:2510.06940 (cs) [Submitted on 8 Oct 2025 (v1), last revised 22 Feb 2026 (this version, v3)] Title:Revisiting Node Affinity Prediction in Temporal Graphs Authors:Or Feldman, Krishna Sri Ipsit Mantri, Moshe Eliasof, Chaim Baskin View a PDF of the paper titled Revisiting Node Affinity Prediction in Temporal Graphs, by Or Feldman and 3 other authors View PDF HTML (experimental) Abstract:Node affinity prediction is a common task that is widely used in temporal graph learning with applications in social and financial networks, recommender systems, and more. Recent works have addressed this task by adapting state-of-the-art dynamic link property prediction models to node affinity prediction. However, simple heuristics, such as Persistent Forecast or Moving Average, outperform these models. In this work, we analyze the challenges in training current Temporal Graph Neural Networks for node affinity prediction and suggest appropriate solutions. Combining the solutions, we develop NAViS - Node Affinity prediction model using Virtual State, by exploiting the equivalence between heuristics and state space models. While promising, training NAViS is non-trivial. Therefore, we further introduce a novel loss function for node affinity prediction. We evaluate NAViS on TGB and show that it outperforms the state-of-the-art, including heuristics. Our source code is available at this https URL Comments: Subjects: Machine Learning (cs.LG) Cite as: arXiv:...
